Choosing RAM & Power Supply
Choosing RAM & Power Supply
I intend to purchase a Ryzen 7 5800x when it arrives and I’ve already ordered a 3090 (still pending). I’m considering adding 32GB of RAM and am unsure which option is superior. I evaluated the Vengeance RGB Pro 32GB DDR4 3200MHz CL16 Dual Channel Kit and the HyperX Fury RGB 32GB DDR4 3200MHz CL16 Dual Channel Kit, both offering identical specifications. No detailed comparison was found. Regarding power, my current 1000W supply should suffice based on the specs, though I remain cautious.
For the RAM, both options match in specifications (and timing is likely the same), so pick the more affordable or visually appealing one. Regarding the power supply, a 1000W unit is sufficient. The 3090 typically handles about 400W, while the 5800X has an estimated TDP of 105W, giving you ample room for overclocking.
Nvidia states the overall power consumption of a 3090 model reaches 750 watts.
Both kits use DDR4-3200 CL16 RAM, likely sharing the same DRAM chips—one from Corsair and one from Kingston. Choose the one that suits your preference or offers better value. For Ryzen processors, opt for higher frequency if budget allows; aim for around DDR4-3600 with Infinity Fabric for full performance. Rumors suggest AMD boosts Ryzen 5000 series to up to 2000 MHz on Infinity Fabric. Regarding power supply, reviews indicate an 850W unit should suffice for the whole system, including a single RTX 3090. With a reliable 850W PSU, you won’t need to upgrade unless you have a 1000W model.
3090 delivers roughly 400 watts, while the 5800X hasn’t been tested yet (it’s still in development), so it’s unlikely to exceed 125 watts.
You're right in your reasoning. Generally, a lower CL (clock speed) is preferable over a higher RAM speed when it comes to performance. A 3200MHz CL16 is typically better than a 3600MHz CL18 because the former offers more stable and consistent performance for most tasks. Memory Express notes that their 3600MHz RAM usually uses a CL18, which supports this trend.
You'll need to perform some calculations to determine the result. It's important to match both frequency and timing requirements. For instance: DDR4-3000 CL14 tends to be slower compared to DDR4-3600 CL16. A simple method to estimate latency in nanoseconds is: multiply frequency by a factor based on pin count and then divide by the clock speed. For example: DDR4-3000 CL14 = 2000 × 14 / 3000 ≈ 9.33 ns DDR4-3600 CL16 = 2000 × 16 / 3600 ≈ 8.88 ns DDR4-3600 CL18 = 2000 × 18 / 3600 = 10 ns DDR4-3200 CL16 = 2000 × 16 / 3200 = 10 ns Many DDR4-3600 CL16 options are available; consider those instead. DDR4-3600 CL18 usually offers lower quality versions of the CL16 models. https://github.com/integralfx/MemTestHel...aster/DDR4 OC Guide.md